On 10 September 2026 the First-tier Tribunal dismissed the appeal in Environmental Services Ltd v HMRC. A waste haulier had claimed SME R&D relief for the periods ended 31 July 2020 and 31 July 2021 on two waste-handling projects (paragraph 26). Its additional yard foreman was accepted as a competent professional, the unwritten plan he carried was held on this fact pattern to be a plan (paragraph 86), and HMRC’s case that the work was merely routine “in the sense of ordinary day-to-day operations” was rejected (paragraph 99). The director who gave evidence was not accepted. The appeal was dismissed regardless (paragraph 111). Our register entry carries the citation, the issues and the holdings.
This article is about the part a director can do something about. The evidence was not the problem; its content was. “[T]he difficulty for ESL is the content of the evidence”, the tribunal wrote of the witness it had just accepted (paragraph 82). At paragraph 90 it named “[t]he principal difficulty”: a company solving the problems its operations and contracts threw up, not running “projects directed towards achieving an advance in science or technology” (paragraph 90). Two tests failed underneath it: no technological uncertainty within the meaning of the Guidelines (paragraph 94), and no baseline against which an advance could be measured (paragraph 95). Quantum then failed on its own feet (paragraphs 102 and 106).
The question we would have asked first
Our own view, plainly. This was skilled engineering under real time pressure, and the tribunal said so: it rejected HMRC’s characterisation of the activities as “merely routine in the sense of ordinary day-to-day operations”, and recorded that its conclusion was not that the company did no innovative work (paragraphs 99 and 101). The decision criticises nobody, and we will not guess at how the claim came to be made.
The question a chartered adviser puts before anything is filed is narrower than that. What could a competent professional in waste-management technology not already do, and what did this work settle for the field rather than for this company? Asked in July 2019, when the M1 contract was won (paragraph 15), it takes an afternoon and a sheet of paper. Asked first in a hearing bundle seven years later, there is nothing left to ask it of. On the evidence recorded in the decision, it was never put to the foreman in writing while the work ran.
Why did the claim fail when the evidence was accepted?
Principally on three findings, with a fourth that stands by itself. The improvements ran to the company’s operations rather than to the field. No technological uncertainty was established. No baseline was articulated. The later period failed separately, on its own thin evidence.
Start with the direction of the benefit, which decides the rest. The tribunal found the projects existed to let the company deliver the M1 contract, and the work to be “largely reactive” (paragraphs 91 and 97). The projects were “directed primarily towards improving ESL’s operational capability and enabling it to service particular customer requirements”, which “is not the same as seeking an advance in overall knowledge or capability in a field of science or technology” (paragraph 98). Commercial pressure makes work happen. It does not make work qualify.
The uncertainty went the same way. The evidence was about “practical difficulties arising from varying waste compositions, changing disposal requirements and operational constraints”. Those are real problems. What they did not establish were “uncertainties concerning what was scientifically or technologically feasible” (paragraph 94). Variable operating conditions, without more, are not a technological uncertainty. Nor is experimentation enough on its own, which the tribunal held after accepting that the work involved it (paragraph 92).
Then the baseline, which is where the gap is named most plainly. “Nor was the Tribunal provided with a clearly articulated technological baseline from which any advance could be measured” (paragraph 95). Neither witness “identified the state of knowledge or capability in the relevant field before the projects commenced”. Neither explained why the solutions were an advance in that field “rather than adaptations of existing techniques and equipment” (paragraph 95).
The fourth finding is the one a director is most likely to need. The company had relied on “the same broad descriptions of the two projects in both periods” (paragraph 96). The tribunal found the evidence for the year ended 31 July 2021 “particularly limited” and “particularly sparse”, and held the burden undischarged for that period alone (paragraphs 96 and 100). Two periods are two claims. A description carried forward is not evidence for this year.
What accepting the foreman was actually worth
He cleared every hurdle put in front of him. The decision calls him an additional yard foreman, thirty years at the company, with City & Guilds and an ONC in motor vehicle engineering (paragraphs 6 and 21). “[A]pproximately 40 years’ practical experience working with heavy vehicles, tanker systems and waste-handling equipment” satisfied the tribunal of his expertise, together with his direct involvement in identifying problems, designing solutions and implementing modifications (paragraph 81). Paragraph 79 holds that the Guidelines mandate no specific qualification. It does not say qualifications were absent here, and a claim reading it as permission to field an unqualified witness has it backwards.
The director was refused. The tribunal agreed with HMRC that she was not a competent professional for the purposes of the Guidelines: her experience lay in running a waste disposal business, and she could not explain the technical detail of the two projects. It recorded in the same breath that she “clearly understands the technical aspects of the projects at a high level and had a detailed understanding of their commercial drivers” (paragraph 80). Understanding the commercial case is not the same as being able to speak for the field, which cost the claim one of its two competent professionals.
What the acceptance bought was less than it looks. Paragraph 82 sets out what his evidence had to cover and did not: the state of knowledge in the field, the uncertainty that state of knowledge threw up, and why the solutions were not ones a competent professional could readily have deduced. What he gave was the practical difficulties, the modifications and the testing. Describing what was done answers none of the three.
What is this the first decision to show?
Two things, both on quantum, and neither is new law; a First-tier Tribunal decision makes none. They are the first published illustrations of either point we have found. Both were reached in the alternative, so the appeal would have failed on the money even had the activities qualified (paragraph 102).
Hired plant is the first, a cost-schedule question rather than a technology one. Even if the projects had qualified, the tribunal agreed with HMRC that the hired equipment, “which from the description appeared to be plant and machinery”, “was not consumed as part of the R&D process, and as such could not qualify for relief as a consumable” (paragraphs 102, 104 and 105). Plant and machinery leases carry “specific tax regimes” of their own, and on the company’s reading “a taxpayer could potentially get tax relief more than once for the same expenditure” (paragraph 105). The tribunal then protected the company from its own reasoning: “[t]here was no suggestion, and we make no inference that ESL was attempting to claim tax relief more than once for the hiring costs” (paragraph 105). That line belongs in the cost schedule, not in an argument at a hearing.
Apportionment is the second, and it will reach more claims. The tribunal would not fix qualifying expenditure on a just and reasonable basis where no evidence showed what was attributable to qualifying activities, and where “the methodology used to derive the claimed percentages of staff time was unclear” (paragraphs 103 and 106). Look at what was in front of it. Eighty per cent of the foreman’s salary and ten per cent of two directors’ salaries for 2020, then fifty and five for 2021 (paragraphs 30 and 31). “We were not provided with any time records or workings” (paragraph 32). A percentage with nothing behind it is a number somebody chose.
Does a First-tier Tribunal decision change the law?
No. First-tier Tribunal decisions bind only the parties and set no precedent. What this one shows is how the tests are being applied now, and it reads across four decisions already in our register, distinguishing none.
It quotes Tanglewood Care Services at length and adopts its reading that “paragraphs 6 and 9 of the BEIS Guidelines should be read together”, so improved functionality alone does not qualify (paragraphs 70 and 88). This is the first application of Tanglewood we have found. The same extract carries the only authority above tribunal level in play, Briggs J in Gripple Ltd v HMRC [2010] EWHC 1609 (Ch), for whom the R&D provisions are “a detailed and prescriptive code” (paragraph 70).
It draws on Flame Tree Publishing without adopting it. Paragraph 74 records the formulation HMRC put to it out of Flame Tree: “appropriate qualifications, experience and up-to-date knowledge of the relevant scientific and technological principles involved”. The formulation was HMRC’s own. It is set out at Flame Tree [66], as counsel’s skeleton argument, and endorsed at [68], where that tribunal had “no hesitation in agreeing with HMRC”; this decision pins it to [68] and attributes it to the tribunal. Paragraph 79 then qualifies it, holding that the Guidelines mandate no specific qualification, and on the facts the tribunal reaches the opposite result.
On Hadee Engineering it goes both ways. It applies the plan requirement at Hadee [217], where the tribunal expected “some record or documentary evidence or, in the absence on which [sic], a detailed explanation” of the uncertainty and how the work was designed to resolve it (paragraph 85). Hadee’s own words are “in the absence of which”. It then accepts an unwritten plan — “Given this fact pattern, we accept that ESL had a method or plan, albeit not one that was written down” (paragraph 86) — and dismisses anyway, because “ESL has failed to explain or evidence that it had a plan which aimed to resolve scientific uncertainties” (paragraph 85). Both sides had agreed to read Guidelines paragraph 9 in the context of paragraph 6, which the tribunal noted mirrors Beer Express (paragraphs 68 and 69).
No onward appeal has been reported as at 18 September 2026, and the window is open: permission must be applied for within 56 days of the decision being sent (paragraph 112). HMRC’s guidance has not moved; the CIRD change log runs only to 25 August 2026. Every decision that shapes an R&D claim has a page in our R&D case-law register.
What this asks of a claim being prepared now
Both periods ended well before April 2024, so this was an old-scheme SME claim on the 2010 edition of the Guidelines. That dates the scheme, not the tests. The same tests sit behind the merged R&D expenditure credit in Chapter 1A of Part 13 CTA 2009 and behind Enhanced R&D Intensive Support in Chapter 2. ERIS is for a loss-making SME whose relevant R&D expenditure is at least 30% of its total relevant expenditure; every other company, including a large R&D-intensive one and a profitable SME, claims the merged scheme. When the period began separates old schemes from current, at periods beginning on or after 1 April 2024, and settles nothing else.
Four things follow, each cheap at the start of a claim and impossible at a hearing.
- Write the baseline down before the work starts. State what is publicly available or readily deducible in the field, in the field’s terms, and date it. The principal finding sits at paragraph 90, and the baseline is what was missing from it. What counts as qualifying R&D takes the four tests in turn.
- Put paragraph 82’s three questions to the competent professional in writing, while the work runs rather than afterwards. Being a competent professional is the start of the evidence, not the end of it.
- Separate the contract from the technology. A customer deadline explains why work happened. It does not establish a scientific or technological uncertainty, and the two read very differently three years later when an HMRC enquiry asks which was which.
- Keep the workings, period by period. A figure for staff time needs the record behind it, and the second year needs its own. That is what the records a claim needs means in practice, and the substance of a claim that withstands scrutiny.
A truthful witness, an accepted competent professional and an accepted plan were not enough. The baseline is the claim. If you cannot tell whether your company’s work clears it, talk it through with a chartered adviser before anything is filed.
Sources
- Environmental Services Ltd v HMRC — [2026] UKFTT 1301 (TC), printed on the face as [2026] UKFTT 01301 (TC); TC 10015; appeal reference TC/2024/02195; heard at Manchester on 17 and 18 June 2026 before Tribunal Member Ann Christian and Tribunal Judge Judith Harrison; released 10 September 2026. Every paragraph number in this article refers to it, verified against the Akoma Ntoso XML.
- Section 1044, section 1051 and section 1052 CTA 2009 as at 31 July 2020 — the old SME scheme provisions in issue. The same text applied at 31 July 2021. All were superseded for accounting periods beginning on or after 1 April 2024.
- Section 1125 CTA 2009, the consumable-items definition behind paragraph 105, and the current version, which adds data licences and cloud computing services but keeps “consumable or transformable materials” — so paragraph 105 reads across to a claim made now.
- Section 1044 CTA 2009 — the current Chapter 2 conditions behind the sentence above: Condition A at subsection (2) is “that the company is a small or medium-sized enterprise in the period”; Condition B at subsection (2A) is that the company “meets the R&D intensity condition in the period” or, under the grace limb at (2A)(b), “obtained relief under this Chapter for its most recent prior accounting period of 12 months’ duration, having met the R&D intensity condition in that period”; and Condition E at subsection (5A) is “that the company makes a loss in the trade in the period”. Section 1045ZA(2) sets the intensity condition at relevant R&D expenditure of “at least 30% of its total relevant expenditure for the period”. Section 1045 carries the same Conditions A and B for pre-trading expenditure. Conditions B and E and section 1045ZA were inserted with effect for accounting periods beginning on or after 1 April 2024.
- Guidelines on the meaning of research and development for tax purposes — paragraph 3 on what R&D is, 6 on the advance, 9 on the examples, 13 on uncertainty, 20 on overall knowledge or capability and 23 on appreciable improvement. The 2004 text as revised in 2010 applied here. The numbers used in this article are the numbers the decision itself uses, at its paragraphs 58 to 73.
- CIRD manual change log — checked 18 September 2026; the most recent entry is 25 August 2026, and none is attributed to this decision. The decision cites no CIRD paragraph.
- Register entries for the decisions read across: Tanglewood Care Services at paragraphs 70 and 88; Beer Express at 69; Flame Tree Publishing at [66], HMRC counsel’s formulation, endorsed by that tribunal at [68]; this decision sets it out at 74, pinning it to [68], and qualifies it at 79; Hadee Engineering at 85 applying Hadee at [217]. Gripple Ltd v HMRC [2010] EWHC 1609 (Ch) at [12] reaches this decision inside the Tanglewood extract at paragraph 70 and has no register entry of its own.
This article describes the rules as they stood at the review date above. The rules change: for the current position, start with our guides or talk to us.
